A Simple Guide to NEOWISE Data Problems Nathan Myhrvold May 25, 2016 In a paper recently submitted to a scientific journal and posted as a preprint on arXiv, I pointed out a number of weaknesses in results previously published – and widely cited – from the NEOWISE asteroid mission. The 110-paGe manuscript is full of detailed arguments about statistics and the physics of how asteroids absorb and reflect sunlight and then emit infrared light. It’s all pretty complicated stuff, written for an audience of professional astronomers, physicists, and statisticians. But the most eGreGious problems with the NEOWISE project are dead simple to explain; indeed, this miGht be a Good project for a middle-school science class. In this guide, I show how anyone can examine the data themselves and spot the most critical errors. It’s a great example of how scientific debates that seem complicated sometimes come down to really simple issues. Background Let’s begin with a little backGround. NEOWISE is a NASA-funded project that spent millions of taxpayer dollars to estimate the diameter of more than 150,000 asteroids, as well as other physical properties of the objects. It did this by analyzinG observations of the infrared (IR) liGht given off by the asteroids and picked up by the WISE satellite. WISE is a space telescope that is a bit like a supersized and more sophisticated version of the thermal cameras used to spot heat leaks in a buildinG. The NEOWISE project used a technique called thermal modelinG to convert the briGhtness of each asteroid at several different infrared “colors” (wavelenGth bands) into an estimate of the object’s diameter. The most widely used thermal model is NEATM (near- Earth asteroid thermal model), which dates from the 1990s and estimates the infrared emission due to the warmth of the asteroid itself. The NEOWISE team of astronomers modified the NEATM model to add infrared emissions due to sunliGht reflected by the asteroid. Accounting for reflected sunlight is more important when analyzing data from the WISE space telescope than it has been for previous telescopes. In addition to estimatinG diameter, the NEOWISE group also estimated other asteroid parameters, such as objects’ albedo, both in visible light and in the infrared. (Albedo is a scientific term for reflectivity – how much of the sun’s liGht reflects from the asteroid.) But the most important parameter is diameter because it effects all of the others. In a series of scientific papers, the NEOWISE team published the diameter and other parameters for about 158,000 asteroids. That’s a huGe contribution. For comparison, the larGest previous study provided data on about 2,200 asteroids. The larGe scope of these studies made them huGely influential. And the data sets have been used by many other asteroid researchers. That is why it is really important to Get the NEOWISE data right. There are at least three other ways than thermal modelinG to estimate the diameter of an asteroid. One is to use radar from larGe radio telescopes to bounce siGnals off the asteroid. It’s a Great technique, but unfortunately it works for relatively few asteroids. A second method is called stellar occultation, which means blocked starliGht. Every now and then (less often than one may think), an asteroid will move in front of a bright star, and observers on Earth see the star wink out for a short period. If you time the event accurately, you can use the duration of the wink to estimate the size of the asteroid. A third and even more accurate method is to visit the asteroid. Some space probes have flown close enouGh to asteroids to get nice pictures from which we can measure their diameters. Recently, for example, NASA obtained wonderful pictures of Ceres, one of the largest asteroids. I refer frequently in this Guide to radar, occultation, and spacecraft analyses, which I lump together with the acronym ROS. Unfortunately, ROS estimates are available for relatively few asteroids. Of the 158,000 asteroids studied by NEOWISE, ROS estimates have been published for only about 150 – so rouGhly one asteroid in a thousand. As a result, the main value of ROS estimates is that they let us check whether the estimates produced by thermal models are accurate or not. The idea is simple. If we can Gain confidence that thermal modelinG works well on the ROS asteroids, then we can feel confident about our understanding of the vastly more numerous asteroids for which all we have are diameters estimated from thermal models. The Problem One biG problem my new paper identifies with certain previous NEOWISE studies is that they didn’t just use ROS diameters as described above – as a way to validate their thermal models. Instead they presented ROS diameters as estimates produced by their thermal models. The diameters were exactly copied. The first paper to consider is (Masiero et al., 2011), which has the title “Main Belt Asteroids with WISE/NEOWISE. I. Preliminary Albedos and Diameters.” Fortunately, The Astrophysical Journal, in A Simple Guide to NEOWISE Problems 2 which the NEOWISE Group published its results, is open access, so you can download the paper here: http://iopscience.iop.orG/article/10.1088/0004-637X/741/2/68. The abstract sums up the purpose of this paper pretty well: “UsinG a NEATM thermal model fittinG routine, we compute diameters for over 100,000 Main Belt asteroids from their IR thermal flux, with errors better than 10%.” Table 1 in the paper presents an excerpt of the most important results, presented at “Thermal Model Fits.” This full table includes more than 100,000 entries and can be downloaded from http://iopscience.iop.orG/0004- 637X/741/2/68/suppdata/apj398969t1_mrt.txt or directly available from Caltech/JPL at http://wise2.ipac.caltech.edu/staff/bauer/NEOWISE_pass1/. Here is a snapshot of Table 1, where I have added some colored graphics to make it easier to see the relevant part of the table. The “Object” columns lists the identifyinG number for each asteroid, and the D column lists the diameter generated by their model – or so they claim. I have drawn a red and green boxes around some of the asteroid entries. In the red boxes, note that the diameters, which are in units of kilometers, all end in “.000.” In contrast, the diameters for asteroids hiGhliGhted by the Green box include non–zero diGits all the way out to the nearest 0.001 km, which is the nearest meter. Note also that some asteroids (such as asteroids 00002 and 00009) appear on multiple rows. That is because the NEOWISE team broke up the data in a non-standard way and performed separate curve fits on 3-day to 10-day seGments rather than simply fittinG all of the data available. It doesn’t make any sense to do that, but that’s a separate and more complicated issue that I discuss in my paper. A Simple Guide to NEOWISE Problems 3 Here is the problem in a nutshell. The diameters boxed in red above – as well as more than 100 others not shown here – are exactly equal, to the nearest meter, to ROS diameters published in papers well before the NEOWISE studies. Some of the supposed NEOWISE results aren’t NEOWISE results at all – they were directly copied from the work of others. You can check this yourself. Asteroid 2 (also known as Pallas or 00002) has a diameter of 544.000 km in the table above from Masiero et al., 2011. Compare it to the entry for that asteroid in Shevchenko and Tedesco, 2006, one of the ROS papers, which is available at http://www.sciencedirect.com/science/article/pii/S001910350600 128X. In Table 1 of the Shevchenko paper we find this: Here, column �"## gives the diameter determined by the occultation method. It matches the Masiero value for asteroid 2 Pallas (aka 00002) exactly. Turning to another ROS source, Durech et al. 2011, available at http://www.sciencedirect.com/science/article/pii/S001910351100 1072 or http://arxiv.orG/abs/1104.4227, we find the followinG table. Check out the value for the diameter of asteroid 5 Astraea (aka 00005): it is 115 km, exactly equal to the diameter listed in the Masiero Table 1 above. You may also notice that asteroid 2 Pallas is also in this table, with a different diameter than in the Shevchenko or Masiero tables. That is typical; one almost never Gets the same number from different ROS studies, mainly as a result of measurement error. You can also see that asteroid 00002 in the Masiero table matches diameter reported by the Shevchenko and Tedesco rather than the estimate Given by Durech et al. It is not clear why the NEOWISE A Simple Guide to NEOWISE Problems 4 authors copied from one source versus another in each case. But it is clear thatthese numbers were copied, not computed as claimed. The excerpt from Masiero Table 1 includes six examples of copyinG for five asteroids. But if you compare the diameters Given in the complete table of more than 100,000 asteroids to those in the ROS references (see links below), you will find, as I did, exact matches in 117 cases, involvinG 102 asteroids. The same problem occurs as well in another one of the main NEOWISE papers: Mainzer, Grav, Bauer, et al., 2011, “NEOWISE Observations of Near Earth Objects: Preliminary Results.” The paper is at http://iopscience.iop.org/article/10.1088/0004- 637X/743/2/156, and its data table is available at http://iopscience.iop.orG/0004- 637X/743/2/156/suppdata/apj408731t1_mrt.txt.